The fourth session of Challenges and Safeguards against AI-Generated Disinformation will discuss Language Model Detectors Are Easily Optimized Against with Charlotte Nicks and Sergey Sanovich on Thursday, November 21, 2024 at 4:00 pm in Annenberg Conference Room, Shultz Building.

Read the paper here

Language Model Detectors Are Easily Optimized Against

ABOUT THE SPEAKERS

Charlotte Nicks is a senior at Stanford, studying mathematics and electrical engineering. She has been selected as a Stanford STEM Fellow and worked at the IRIS Lab at the Stanford Artificial Intelligence Laboratory on AI detection. Her academic interests include signal processing, machine learning, and complex analysis. She has served as co-President of Affiliated Stanford Entrepreneurial Students, Stanford’s global entrepreneurship society, and Stanford Students in Biodesign. During summers, Charlotte interned as a quantitative trader at Jane Street and as an autonomy engineer at AeroVironment.

Sergey Sanovich is a Hoover Fellow at the Hoover Institution. Before joining the Hoover Institution, Sergey Sanovich was a postdoctoral research associate at the Center for Information Technology Policy at Princeton University. Sanovich received his PhD in political science from New York University and continues his affiliation with its Center for Social Media and Politics. His research is focused on disinformation and social media platform governance; online censorship and propaganda by authoritarian regimes; and elections and partisanship in information autocracies. His work has been published at the American Political Science ReviewComparative PoliticsResearch & Politics, and Big Data, and as a lead chapter in an edited volume on disinformation from Oxford University Press. Sanovich has also contributed to several policy reports, particularly focusing on protection from disinformation, including “Securing American Elections,” issued by the Stanford Cyber Policy Center at its launch.


ABOUT THE SERIES

Distinguishing between human- and AI-generated content is already an important enough problem in multiple domains – from social media moderation to education – that there is a quickly growing body of empirical research on AI detection and an equally quickly growing industry of its non/commercial applications. But will current tools survive the next generation of LLMs, including open models and those focused specifically on bypassing detection? What about the generation after that? Cutting-edge research, as well as presentations from leading industry professionals, in this series will clarify the limits of detection in the medium- and long-term and help identify the optimal points and types of policy intervention. This series is organized by Sergey Sanovich.

Upcoming Events

Thursday, May 22, 2025
Hoover Tower Rotunda
Snippets From The 20th Care And Conservation Of Manuscripts Conference, Copenhagen 2025
The Hoover Institution Library & Archives invite you to a talk with book and paper conservator Laurent Cruveillier about Snippets from the 20th… Stanford, California
Friday, May 30, 2025
america
More Than Red Vs. Blue: Stories, Struggles, And Strengths In Rural America
The Center for Revitalizing American Institutions (RAI) hosts More Than Red vs. Blue: Stories, Struggles, and Strengths in Rural America on May 30,… Hoover Institution, Stanford University
Tuesday, June 3, 2025
The Party's Interests Come First: The Life of Xi Zhongxun, Father of Xi Jinping
The Party's Interests Come First: The Life Of Xi Zhongxun, Father Of Xi Jinping
The Hoover History Lab and The Hoover Institution Library & Archives invite you to The Party's Interests Come First: The Life of Xi Zhongxun,… Shultz Auditorium, George P. Shultz Building
overlay image